In order to increase the safety of all air travel, technologies that continueto augment the pilot's ability to avoid collisions and stay clear of danger areneeded. But, before these can be certified and deployed, their performance andpotential failure cases have to be understood. This requires evaluating a modelof the system on simulated encounters, consisting of different trajectoriesthat should replicate the real world. This is commonly done using a statistical encounter model, which produces largeamounts of data but relies on the accuracy of the statistical model, thuslimited in its ability to produce realistic data. The goal with this project isto create an encounter dataset of real trajectories that would provide analternative to encounter models. This is done using an ADS-B dataset from The OpenSky Network (provided byDaedalean AI), consisting of 226 billion air traffic data points from 2019.First, a solution to efficiently query and reconstruct trajectories from thedataset is designed and implemented. Using it, a NMAC (Near Mid-Air Collision)dataset is created to demonstrate the viability of ADS-B as a source forcreating an encounter dataset, and to prove the capabilities of the designedsolution.
Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:kth-331369 |
Date | January 2023 |
Creators | Hermansson, Leopold |
Publisher | KTH, Skolan för teknikvetenskap (SCI) |
Source Sets | DiVA Archive at Upsalla University |
Language | English |
Detected Language | English |
Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
Format | application/pdf |
Rights | info:eu-repo/semantics/openAccess |
Relation | TRITA-SCI-GRU ; 2023:179 |
Page generated in 0.0106 seconds